Counterparty Credit Risk: Definition, a Swap on Real Rates and How Netting Shrinks It

Michael Whitman · Trading since 2017, full-time since 2020 · YouTube creator, 137K subscribers · Published 2026-09-14 · Updated 2026-11-02 · Reviewed 2026-11-01

Counterparty credit risk is the risk that the other party to a derivative or financing trade defaults while the trade is worth money to you, before it has paid what it owes. The amount at risk is the cost of replacing the contract at that day’s prices, so it is unknown when you sign and can end up on either side.

A bank loan is the easy case: lend $10 million and $10 million is what you can lose. Sign a swap and the number you can lose is zero on the first day, then grows, shrinks or switches sides as rates move. The swap below, valued on real rates, shows how far it can travel.

How it works

A derivative is a promise to pay later. Between the trade date and the last payment, either side can fail.

Exposure is replacement cost. If the counterparty vanishes, you have to strike the same deal with someone else at today’s prices. The difference between the old terms and the new ones is what you lose.

At the start it is about zero. A swap struck at the market rate is worth nothing to either side on day one. Exposure appears only because prices moved after the handshake.

And it runs both ways. Whoever the contract currently favors holds the exposure to the other. When rates turn, the exposure changes hands.

A single fund can lean on many dealers at once; TQQQ held its swaps with 8 to 10 banks in every filing, as the total return swap page shows.

The four numbers risk teams use

Current exposure is the replacement cost today, if positive, and zero if the contract is worth less than nothing to you, because then you owe them.

Expected exposure (EE) is the average of that positive value at a future date, across many simulated paths for prices.

Potential future exposure (PFE) is a high percentile of the same distribution, often the 95th: a bad-but-plausible size the exposure could reach before the contract ends.

Credit valuation adjustment (CVA) turns exposure into a price. In its simplest form it is loss given default times expected exposure times the chance of default, added up over the life of the trade.

A worked example: a $100 million swap on five years of real rates

The trade. On 1 November 2021, party A agrees to pay a fixed 1.58% a year on $100 million for ten years, and party B agrees to pay a floating rate. 1.58% was the 10-year Treasury yield that day, so on this page’s method the swap starts at a value of $0 to both.

How we value it. Each month-end, the fair fixed rate for the years left is read off the 5-, 7- and 10-year Treasury yields (straight-line between them), and the swap’s value to A is $100 million times (that rate minus 1.58%) times an annuity factor for the years left, discounted at that same rate with annual payments. Treasury yields stand in for swap rates here. Real swaps are marked off swap rates, which differ from Treasury yields by a gap called the swap spread, so these are estimates, not dealer marks. Every month is in the swap valuation file.

Month one, the exposure belongs to B. By 30 November 2021 the rate for 9.92 years had dipped to 1.43%. A was paying more than the market rate, so the swap was worth -$1.40 million to A and B held a $1.40 million exposure to A.

Then it flips. B still held $0.58 million of exposure at the end of December. Then rates climbed through 2022. By 31 January 2022 the value to A was $1.83 million, and it stayed above zero at every later month-end to October 2026.

The peak. On 31 October 2023, with 8.00 years left, the fair rate was 4.89% against A’s 1.58%. With an annuity factor of 6.50, the swap was worth $21.48 million to A. If B had failed that day, A would have had to pay up to strike a replacement at 4.89%, and it would have stood in B’s insolvency as a creditor for roughly that $21.48 million.

Where it stands. On 29 October 2026, with 5.01 years left, the fair rate was 4.78% and the annuity factor 4.36, so the value was $13.96 million. Same contract, same $100 million notional, and the amount at risk had moved from $0 to $21.48 million and back to $13.96 million without either party doing anything.

What a CVA would charge for it. As an illustration only: if B had a 1% chance of defaulting over the next year and A expected to lose 60% of the claim, the one-year expected loss on today’s $13.96 million would be about $83,776. Real desks run this over every future date with simulated rates, not one number.

Why it is harder than lending risk

The exposure is bilateral. In a loan only the lender is exposed. In the swap above, B was exposed in the first two months and A in every month after that, so each had to vet the other before signing.

And it has to be estimated across the contract’s life. On 1 November 2021 nobody could have known the exposure would reach $21.48 million. That is why the industry sizes limits from PFE, a percentile of what the swap could be worth, rather than today’s value.

Wrong-way risk

The dangerous case is exposure that grows as the counterparty weakens. If the thing pushing your counterparty toward failure is the same thing making the contract valuable to you, the two risks are not independent.

A concrete shape. Buying protection against a company’s default from a bank heavily exposed to that same company means the protection is worth most when the seller is least able to pay.

Or a currency example. A swap that pays you when a country’s currency collapses, transacted with a bank in that country, is a claim that becomes large at the moment the institution behind it is under maximum strain.

Which is why the discipline treats wrong-way risk separately. Ordinary exposure is often modeled as if market moves and default were unrelated; wrong-way risk breaks that assumption.

The original data

One swap shows the mechanism; the Bank for International Settlements shows the scale. It collects derivatives positions from the large dealers in its reporting countries every half-year. Its bulk file, downloaded on 27 October 2026, runs from June 1998 to end-December 2025. Three of its measures, all counterparties and all risk categories:

Notional is the size of the contracts, not the money at risk. In the swap example that is the $100 million. Gross market value is what it would cost to replace every contract at that day’s prices, the $13.96 million kind of number, counting contracts in a dealer’s favor and against it. Gross credit exposure is that value after legally enforceable bilateral netting and before any collateral. So in December 2025, netting took 85.30% out of gross market value.

The finding: netting removes far more than it used to. In June 1998 credit exposure was 46.95% of gross market value. It has been under 20% in 27 of the 56 half-years: both halves of 2005, every half-year from 2008 to mid-2016, and every one since 2022. Its low was 13.58% in June 2025. At the end-2008 peak, gross market value hit $34.94 trillion while exposure after netting was $5.01 trillion, 14.32%.

Clearing is the other half of the change. Central counterparties held 78.75% of interest rate derivatives notional at end-2025, against 75.76% in the first half of 2016, when the series begins.

The check: the OCC’s report on insured US banks for the fourth quarter of 2025 says netting agreements “allowed banks to reduce GPFV exposures by 89.1 percent”, leaving $241 billion of net current credit exposure from $2,212 billion of gross positive fair value. That counts only contracts in the banks' favor, a narrower base than BIS’s, so the two percentages measure the same effect rather than the same number. Both are in the netting check file and the half-yearly series in the BIS series file. The set of reporting dealers and the netting law behind the figures have changed over 27 years, so the early ratios are not a like-for-like sample.

How it is actually managed

Netting agreements. A master agreement covering many trades means only the net amount is owed on default. If A also had a second swap with B worth -$5 million to A, the claim on B would be the difference, not the larger figure.

Collateral. Exposure above a threshold is covered by assets posted daily, which converts credit risk into liquidity risk and operational risk — better risks, and not no risk. Under a daily margin agreement, B would have been posting cash to A all through 2022 as the swap’s value climbed.

Central clearing. Standardized contracts are novated to a clearing house, which replaces the counterparty entirely.

And pricing. The cost of counterparty credit is charged into the trade as a valuation adjustment, so a weaker counterparty receives a worse price rather than being refused.

The valuation adjustments

CVA is the credit valuation adjustment - the expected loss from the counterparty defaulting, priced into the trade and charged to the client rather than absorbed quietly.

DVA is the same calculation pointing the other way, the value of your own possible default to you, which is correct in theory and deeply uncomfortable in practice: a firm’s own deterioration creates an accounting gain.

And FVA covers the funding cost of the collateral an uncollateralized trade obliges the dealer to post elsewhere, which is a real expense somebody has to carry.

When it fails

The characteristic failure is netting that does not hold up. A firm calculates its exposure on the assumption that gains and losses across hundreds of contracts with one counterparty will offset on default. That offset depends on a master agreement being enforceable in the counterparty’s jurisdiction, in insolvency, against an administrator with incentives of their own. Where the legal opinion turns out to be optimistic, the net exposure that was carefully managed becomes a gross exposure that was never sized for. On the BIS end-2025 figures, that is the gap between $3.35 trillion and $22.80 trillion.

A second failure is sizing limits from today’s value. In November 2021 the swap above showed A no exposure at all; two years later it was $21.48 million.

A third is treating collateral as complete protection. It is revalued with a lag, and a gap move happens between valuations.

A fourth is ignoring wrong-way risk, where exposure and default probability rise together, and modeling potential exposure from calm data, which understates the tail.

And a fifth is forgetting it is mutual. Your counterparty is running the same analysis on you, and your own deterioration changes what they will trade. The same idea at retail scale is counterparty risk with a broker, and in its narrowest form it is settlement risk, the window when one side has paid and the other has not.

Questions people ask

What is a counterparty risk example?

An interest rate swap. The $100 million swap struck at 1.58% in November 2021 was worth an estimated $21.48 million to the fixed payer by October 2023. Had the floating payer defaulted then, that was roughly the loss before any collateral, on this Treasury-based estimate.

How is CVA calculated?

In its simplest form: loss given default times expected exposure times probability of default, summed over future dates. With a 1% one-year default chance, 60% loss given default and $13.96 million of exposure, one year’s piece is about $83,776.

What is the greatest source of counterparty risk?

Over-the-counter derivatives, by size. BIS put dealers’ gross market value at $22.80 trillion at end-2025, before netting cut the credit exposure to $3.35 trillion.

What is the difference between counterparty credit risk and credit risk?

Credit risk on a loan has a fixed amount at risk, owed one way. Counterparty credit risk has an amount that moves with prices and can belong to either side.

Is notional amount the same as the amount at risk?

No. At end-2025 the $844.58 trillion of notional outstanding carried $22.80 trillion of gross market value and $3.35 trillion of credit exposure after netting, 0.40% of notional, before collateral.

Related

Counterparty risk covers the general exposure to the other side. The clearing house page explains the institution set up to take this risk off both sides. And interest rate swap covers the contract type in the worked example above.

This page is educational, not financial advice. Test every idea on your own charts before risking money.